UN Population Division, Department of Economic and Social Affairs,
with support from the UN Population Fund (UNFPA)
|
| UN Population Division, Department of Economic and Social Affairs, with support from the UN Population Fund (UNFPA) |
|
*************************************************************************
For further information please contact the United Nations Population Fund
at 220 East 42nd Street, New York, NY 10017 USA or via E-mail:
vlassoff@unfpa.org
*************************************************************************
FROM: Population Growth and Economic Development: Report on the
Consultative Meeting of Economists Convened by the United Nations
Population Fund, 28-29 September 1992, New York. New York:UNFPA, 1993.
RECENT DEVELOPMENTS IN RESEARCH INTO THE RELATIONSHIP
BETWEEN POPULATION GROWTH
AND ECONOMIC DEVELOPMENT
Background Paper Prepared by UNFPA
I. Introduction
Two contrasting schools of thought are discernible in post-World
War II research in the area of population and development. The first,
which one reviewer calls "orthodoxy", arose in the 1950s and is typi-
fied by the overall conclusion of the first U.S. National Academy of
Sciences report on population and development (National Research
Council, 1971): rapid population growth poses a serious, even in-
surmountable barrier to economic growth and social advancement. The
authors of the report summarized their conclusions as follows:
"Rapid population growth slows down the growth of per capita
incomes in less developed countries and tends to perpetuate
inequalities of income distribution. It holds down the
level of savings and capital investment in the means of
production and thereby limits the rate of growth of gross
domestic product. Food supplies and agricultural production
must be greatly increased to meet the needs of rapidly
growing populations, and this constrains the allocation of
resources to other economic and social sectors. The number
of persons entering the labour force grows very rapidly.
Because the number of people seeking employment is larger
than the number of available jobs, unemployment and
underemployment are increasingly serious problems. An ever
larger number of workers cannot be absorbed into the modern
(industrialized) sector. They are forced into unproductive
service occupations or back into the traditional
(agricultural) sector with its low productivity and bare
subsistence wage levels. ... Widespread poverty, the low
productivity of labour, the growing demands for food, and
slow industrialization distort and degrade the international
trade of the less developed countries."
The World Bank's World Development Report 1984 also embodied
this "orthodox" view. Other major research surveys that concluded
that population growth tends to slow development were those by R.
Cassen (1973), N. Birdsall (1977) and G. McNicoll (1984).
The second tendency, "revisionism", arose in the 1980s and is
typified by overall findings contained in the second NAS report
(National Research Council, 1986), namely, that rapid population
growth is only a secondary factor in economic development, and that
its deleterious effects will largely (though not necessarily
entirely) be countered by market reactions if markets are allowed to
function freely. A statement of the revisionist view is given by A.
Kelley (1988, p. 1715) in concluding his review of research in the
area of population and development:
"Economic growth (as measured by per capita output) in many
developing countries would have been more rapid in an
environment of slower population growth, although in a number
of countries the impact of population was probably negligible,
and in some it may have been positive. ... Because there is no
believable and generally accepted quantitative estimate of
population's effect on development, only a qualitative (a
direction-of-impact) assessment can be made. The assessment,
positive or negative, varies from country to country, over
time, and possibly with the rate of population growth. What
is clear is that an assessment of the impact of population
growth on economic development is highly complex, that
problems like unemployment, famine, and malnutrition are
caused by many factors (including rapid population growth),
and that an emphasis on policies of slowing population growth
without simultaneously confronting the other fundamental
causes of such problems may well lead to disappointing
results."
This marked a return to the earlier non-committal tone of
researchers such as Kuznets (1965) and R. Easterlin (1967), which has
been echoed in the declarations of the international population
conferences convened by the United Nations at 10-year intervals since
1954.
In this review, the focus is on research in the area of
population and development since the 1986 NAS report, the latest
major review of the subject. In cases where post-1986 research has
not modified the 1986 report because of the lack of new research
initiatives, this fact is simply noted. In these cases the
interested reader can find detailed citations either in the NAS
report or in the accompanying background papers (G. Johnson and R.
Lee, 1987).
Given this approach, it is worthwhile to sketch briefly the
major policy conclusions of the 1986 NAS review with respect to the
different effects of population growth:
Per worker capital, output and consumption: Slower population
growth has a net positive effect on the capital/labour ratio; and a
probable positive effect on the savings rate. While "capital
deepening" is a genuine consequence of reduced population growth, its
effect "may be relatively modest".
Technological innovation and economies of scale: In manufac-
turing, this is an urban question; economies of scale are exhausted
at moderate city size. Therefore, slower population growth has no
negative effect on productivity in manufacturing. In agriculture,
population density increases are positive (choice of technique,
infrastructure economies of scale). But this does not imply that net
productivity increases. Some evidence exists that after reaching 100
persons/sq km, overall productivity starts to fall due to diminishing
returns.
Per capita levels of schooling and health: In education, slower
population growth is associated with higher expenditure per pupil;
the same is true of per capita spending on health and nutrition
within households. In government health programmes, it is "unlikely
that rapid population growth is a major impediment" to their success.
Inequality in the distribution of income: In the short term,
slower population growth decreases inequality of income if family
planning is targeted at low-income groups. In the long term there
should also be a lessening of inequality, since labour is favoured
over other factors of production.
Absorption of workers into the modern sector; problems of urban
growth: Population growth has exacerbated some urban problems
(straining services and slowing the growth of the modern sector,
e.g.), but slower population growth "will probably not, by itself,
solve these problems". An urban bias of government policies is a
much more important cause of urban problems than population growth.
Externalities: Some externalities are the result of
population growth: congestion, over-rapid resource degradation, and
inter-generational costs, e.g. It is likely that there are net
negative externalities of child-bearing in most developing
countries. A minimum policy response should include subsidized
family planning programmes; a more ambitious response would involve
"alterations in incentive structures".
The conclusions of the 1986 NAS review are "revisionist", while
still pointing out many aspects of economic development that would
benefit from slower population growth and the population policies
that lead to slower growth. The present review of post-1986
research, while not suggesting radical departures from the NAS
findings, leads to a somewhat stronger conclusion regarding the
negative effects of population growth on development prospects in
developing countries.
II. General Considerations
The main conclusion that emerges from the neo-classical one-
sector model of economic growth is that an increase in the rate of
population growth will lead to a reduction in the long-term
equilibrium level of output per capita. Economists have long
expressed concern, however, that demographic factors may be
inadequately reflected in the simple model. For example, the rate
of population growth and resulting changes in the population's age
structure might change the savings rate, a crucial model parameter.
The rate of technical progress, taken as exogenous in the simple
model, might respond to population factors. The negative effects of
more rapid population growth might be ameliorated by economies of
scale arising from larger population size; on the other hand, labour
productivity might suffer if rapid population growth impedes the
formation of human capital.
The organization of this paper reflects this neo-classical
tradition of looking at the effects of rapid population growth in
terms of the sources of growth: accumulation of physical capital
(i.e., savings and investment); labour-force growth and absorption,
including the shift of labour from low- to high-productivity sectors;
the quality of human resources, which is closely related to the
accumulation of human cap-ital; and economies of scale and technical
and institutional change. Other sections look at research into the
effects of population growth on externalities and on income
distribution and poverty; a final section reviews the results of
recent correlation studies and modelling exercises.
The temporal focus is on the medium-term, that is, roughly 10
to 15 years. This is long enough for changes in fertility and
mortality to cause perceptible shifts in population size and age
structure, but short enough to be of practical interest to policy
makers.
It should be kept in mind that the effects of population growth
are likely to vary widely with national context. Among the
significant background factors may be population density, the
openness and trade orientation of the economy, and the nature and
role of the agricultural sector and the natural-resource base.
Perhaps most important of all is the question of whether local
conditions and institutions are conducive to the allocation of
economic resources through neo-classical market mechanisms.
III. Rapid Population Growth and Savings and Investment
The seminal work in the area of population growth and savings
is that of A. Coale and E. Hoover (1958), who reached an unequi-vocal
conclusion: "To postpone fertility reductions in low-income countries
is to shrink the potential growth in per capita income for the
indefinite future." These researchers hypothesized that fertility
reduction would have four effects:
(1) A straightforward denominator effect. A drop in
fertility would lead to fewer adult-equivalent
consumers in the population. This denominator
effect may, in turn, be decomposed into rate-of-
growth and age-structure effects. Both effects
arise from simple accounting identities and not from
any behavioural model.
(2) A household-level savings-rate effect. The ratio of
young net consumers to older net producers would
fall, resulting in the saving of a higher proportion
of household income.
(3) A national-level composition-of-investment effect.
The reduction in the ratio of children to adults
would allow the state to direct investment away from
schools and hospitals and into directly productive
plant and equipment. (Schools and hospitals,
whatever their beneficial effect on the quality of
the human-resource base over the long term, do not
contribute directly or immediately to production.)
(4) A female labour-force participation effect. In the
lower-fertility scenario, women would supply more
labour to the marketplace. This allowed Coale and
Hoover to assume that, at least for the first 30
years, lower fertility would make no significant
difference in the size of the labour force. In
point of fact, the importance of the labour-force
participation rate effect is marginal in the
simulations presented.
As the behavioural effects (2)-(4) affect the level of GDP in
the numerator of the chosen welfare index, income per adult-
equivalent consumer, we may refer to them as "numerator effects".
As Coale (1986) himself made clear, assumptions (2) and (3) þ which
have generated much debate and refinement þ were less important, in
terms of their contribution to the result of their simulation
analysis, than was (1), the simple denominator effect. Only one-
quarter of the increase in income per adult-equivalent consumer found
to arise from fertility reduction was due to numerator effects; the
rest was, essentially, pure accounting.
The household savings-rate effect (as well as the more complex
question of whether such an effect, if it exists, reduces the
aggregate supply of savings) has been among the most active areas of
research in the field of population and development. Despite the
large volume of research, however, not much has been found in the way
of hard conclusions. H. Leibenstein (1976, p. 618) wrote almost 20
years ago that the issue of whether population growth reduced the
availability of savings could not be answered based on then-available
research. McNicoll (1984, p. 206) concluded that "in general, rather
little" can be concluded on the savings-investment effect of rapid
population growth, and suggested that the phrase "based on presently
available research" could be deleted from Leibenstein's sentence.
A. Mason (1988) summarized his survey of research in the area as
follows:
"A succinct and uncritical reading of the evidence might be
summarized as follows: A high rate of saving is required to
maintain a level of investment necessary for rapid economic
growth. And in many countries, reduced population growth will
facilitate efforts to achieve higher rates of saving. Providing
accurate, rather than succinct, answers to these questions is
more difficult because the processes involved are complex and
because the circumstances vary widely from country to country."
The sources of theoretical difficulty are several. At the
household level, even a simple accounting model of saving behaviour
reveals that birth order, household size and income may influence a
family's reaction to a marginal birth. At the macroeconomic and
population-wide level, there are important age-composition effects
þ the ratio of young households accumulating savings to older
households drawing down savings is a crucial variable þ as well as
an interaction effect between the dependency ratio and the rate of
macroeconomic growth. Because of this interaction, the burden posed
by population growth should be most significant in dynamic, rapidly
growing economies.
Perhaps not surprisingly, the empirical relationship between
aggregate saving rates and the demographic characteristics of
populations is almost as problematic as that between population
growth and per capita economic growth (see the discussion in Section
IX). The basic article supporting an important dependency-ratio
effect is that of N. Leff (1969), which has been much criticized and
often rebutted (see Mason, 1988, for citations). The most-recent
regression analysis of the relationship between aggregate saving
rates and dependency rates leaves the reader with little confidence
that the matter will ever be resolved: parameter estimates are
insignificant and erratic (L. Schumaker and R. Clark, 1992). The
methodology used in this study, however, has been severely criticized
(see above). A model constructed by Mason (1987) suggested that even
a massive shift in the dependency ratio (population under age 15
divided by population over 15) from 0.4 to 0.8 might reduce the
saving rate in a typical developing country only from 22 to 19 per
cent.
A 1987 investigation by the United Nations Conference on Trade
and Development, on the relationship between population and
savings/consumption, showed that, in Latin America, dependency-ratio
effects are statistically significant predictors of savings
behaviour. In areas with high population growth and young
populations, substantial rises in consumption occur as dependency
increases.
Finally, the literature on population growth and savings must
be placed in a broader sources-of-growth context. Studies have,
without exception, assigned a surprisingly small share of long-run
growth to accumulation of physical capital. The most-often cited of
these is that of E. Denison (1985), who apportioned U.S. economic
growth between 1929 and 1981 as follows: 25 per cent to labour-force
growth, holding education level and the distribution of labour force
across sectors constant; 16 per cent to the rising education level
of the labour force; 11 per cent to re-allocation of labour from low-
to high-productivity sectors; 12 per cent to capital growth; 11 per
cent to the reaping of economies of scale; 34 per cent to technical
progress; and -9 per cent to errors and omitted factors.
IV. Rapid Population Growth and Labour-Force Growth and Absorption
Demographic increase has both accounting and behavioural effects
on labour supply. The former is straightforward: a person born today
will represent a job-seeker some 15 to 20 years hence, subject to
prevailing rates of survival, school attendance and labour-force
participation. A particular characteristic of rapidly growing
populations is the preponderance of young persons of limited
experience. This may tend to diminish productivity and wages; on the
other hand, H. Leibenstein (1967) has suggested that the attitudes
of such workers may be particularly conducive to dynamic change, the
adoption of new technologies, etc. (see discussion of labour-embodied
technical change in Section VI).
Little work has been done since the 1986 NAS report on the
effect of fertility and household size on labour supply at the
household level. It has long been known (e.g., G. Standing, 1978)
that high fertility is correlated with low female labour-force
participation in the modern sector and high female labour-force
participation in the agricultural and informal sectors. Casual
observation indicates that high fertility is associated with the
participation of older children in the "household economy" þ
gardening, tending younger siblings, etc. þ which may, as one of its
effects, free up adults for extra-mural economic pursuits.
The "Chayanov effect" states that an extra birth may increase
the labour supply of the head of household, but this hypothesis has
never been tested in a developing country. It is worth remembering
that the male peasants studied by Chayanov were working something
like 700-800 hours per year on average; i.e., they were engaged in
highly seasonal agricultural activities which left free substantial
blocks of leisure time. Such conditions may be replicated in some
agricultural settings, but are unlikely to prevail in an urban
context.
At the national level, studies have revealed no correlation
between the rate of population growth and the rate of unemployment;
on the other hand, open unemployment rates are a notoriously
inadequate measure of labour-market conditions in developing
countries. Labour-force entrants who cannot be absorbed into the
modern sector will shift their labour supply to the household economy
and the informal sector, which act, in effect, as residual sectors.
They may also, if possible, defer searching for jobs and remain in
school.
The major survey of labour markets and employment in developing
countries is that of D. Bloom and R. Freeman (1987). While finding
scattered empirical support for the proposition that rapid population
growth lowers wages, these researchers admitted that the great
variation in local labour markets made it impossible to reach any
strong general conclusion. Some developing countries have succeeded
in absorbing their rapidly growing populations into productive
employment, but this is not necessarily grounds for optimism.
According to dual labour market theory, wages are rigid in the formal
sector but flexible in the informal sector. The effect of rapid
population growth should, therefore, be to (1) increase the absolute
number and proportion of workers in the informal sector, (2) drive
down wages in the informal sector, and thus (3) widen the gap between
formal and informal sectors and worsen the distribution of income
(H.R Hemmer and C. Mannel, 1989; T. Chaudhuri, 1989).
Informal-sector employment is productive and may result in the
accumulation of skills. On the other hand, there exists a hierarchy
of activities, running from the highly desirable to the most
marginal, within the informal sector itself (W. Cole and B. Fayissa,
1991). It may be the case that the "elite" sub-sector of the
informal sector þ auto repair, air-conditioner and refrigerator
maintenance and the like þ is only slightly more flexible than the
modern sector, in which case rapid labour-force growth simply pushes
more workers down the productivity ladder into marginal activities.
According to neo-classical theory, rapid labour-force growth,
by reducing the wage rate relative to the rate of return on capital
(see the discussion in Section VIII), should promote the adoption of
labour-intensive techniques of production. In agriculture (which is
outside the scope of this survey) there is a large body of scholar-
ship on how cultivators have reacted to population pressure by
adopting more labour-intensive production techniques.
There is little question that the neo-classical choice-of-
technique mechanism also flourishes in the informal sector, where
capital intensity is low and elasticity of substitution high.
Researchers have questioned the relevance of neo-classical factor
substitution in the formal sector, however. Here production
techniques may be: imported with little regard to local factor
availabilities; influenced by industrialists' "engineering bias"; or
bound to internationally accepted standards of quality control which
require capital-intensive production techniques (N. Egea, 1990; H.
Pack and L. Westphal, 1986).
V. Rapid Population Growth and Human-Resource Quality at the
Household and National Levels
As empirical work has failed to assign a dominant role to
physical capital formation, there has been a correspondingly strong
interest in human capital and the quality of the human-resource base.
In contrast to the ambiguous evidence regarding the effect of
demographic parameters on physical capital formation, there is rather
strong evidence that high fertility, close birth spacing and large
family size tend to reduce human capital formation by impairing child
development.
High fertility and unfavourable child development can co-occur
in one of three ways: (1) unfavourable overall economic conditions
may cause both; (2) high fertility may impede child development
directly (i.e., holding resources per child constant þ presumably the
impairment occurs via unspecified biological mechanisms or as a
result of reduced parental attention per child); or (3) indirectly
by reducing endowments of nutrition and education per child.
Many studies have found statistically significant associations.
Cross-household surveys have revealed an inverse relationship between
the number of offspring in a family and per-child educational
expenditure; there is also evidence that unwanted births, which are
likely to be higher-parity births, are associated with lower school
enrolment rates for all children in the family. A strong inverse
empirical relationship has been found between the number of children
in the household and child nutrition.
Though studies have not shown that large family size is
associated with low infant- and child-survival rates, there are
indications that large family size is inversely associated with
children's health. Large family size is generally associated with
short birth intervals, and many studies have concluded that closely
spaced children are subject to higher mortality. There are
substantial reductions in infant mortality to be gained from wider
birth spacing (see J. Potter, 1991, pp. 224-30, for references).
Evidence of causality in the area of fertility and child
development has been surveyed by M. Rosenzweig (1988 and 1990), who
drew attention to the deficiencies of available data in the area.
Relatively few studies have been able to go beyond correlations,
which may over-estimate the strength of hypothesized relationships.
Reviewing those which have done so, Rosenzweig (1988, p. 83)
concludes on a cautionary note: "Each assertion [points (1)-(3)
above] can be supported. But the quantitative evidence for any one
is not overwhelming; there are only a few studies in each case that
go beyond correlations, and the estimated magnitudes of the causal
relationships are small."
Data problems have also hampered researchers' ability to
establish the common-sense proposition that adverse childhood
conditions result in lowered adult productivity and earnings. The
individual and household-level evidence cited by G. Cornia (1989,
especially pp.179-80) is scattered and particularistic, but at least
consistent. Birdsall and C. Griffin (1988, pp. 49-50) found in their
survey "limited evidence" from household studies that the reduction
in resources per child associated with high fertility had adverse
long-run effects. They concluded, "The probable (though not well-
charted) result is slower per capita income growth."
Rapid population growth puts an undoubted strain on national
resources in the areas of education (Najafizadeh and Mennerick, 1988;
G. Jones, 1990) and health (Jones, 1990). Some cross-sectional
research has found that national indices of health and education are
positively associated with the rate of economic growth (E. Scholing
and V. Timmermann, 1988). I. Otani and D. Villanueva (1990) found
a significant positive growth effect of the share of government
expenditure on human-resource development, albeit less so than the
traditional "hard" variables such as the saving rate and the rate of
growth of exports. Neither of these studies goes, to use Rosenzweig's
terminology, "beyond correlations".
The lagged Physical Quality of Life Index (PQLI), which should
be a fair index of "basic needs", shows a far stronger correlation
with current-period GDP per capita than does lagged GDP per capita
with the current-period PQLI (B. Newman and R. Thompson, 1989). The
simulation model of D. Wheeler (1984), which indicated that invest-
ment in education, nutrition and family planning could be more
productive in the long run than investment in physical capital, is
still highly regarded and often cited (see also the discussion in
Section IX).
J. Behrman (1990) examined the contribution of training and
indigenous research and development to economic growth and found
"surprisingly little systematic quantitative evidence" (pp. 89-90)
of a causal relationship at the national level. In the area of
health, there is no consistent statistical relationship between
aggregate health spending and measures of health outcomes such as the
infant mortality rate and life expectancy. There seems little doubt
that in most countries, demographic dilution is a secondary factor
when the mis-allocation of resources þ to secondary and post-
secondary education at the expense of primary education; to urban
hospital-based health care at the expense of primary health care in
rural areas and on the urban fringe, etc. þ is taken into account.
G. Becker (1988) elaborated a neo-classical growth model in
which fertility and endowment of children with human capital are both
endogenous decisions, and possibilities of inter-generational
transfers exist. The model's solution is characterized by two stable
equilibria: one with large family size and low human capital per
child, the other with small family size and high endowment of human
capital per child. The practical implication is that some exogenous
push may be required to displace low-income countries away from the
first of these steady states (low-level trap) to the second.
VI. Rapid Population Growth, Economies of Scale and Induced
Technical and Institutional Change
In the years leading up to 1986, a large volume of literature
addressed the relationship between population growth and economies
of scale and induced technical and institutional change; McNicoll
(1984) provides a good survey. This was occasioned, in part, by
publication of The Ultimate Resource (J. Simon, 1981), which argued
that population growth has a strong positive effect on economic
growth. Formal mechanisms by which density-dependent technical
change may lead to escape from near-term Malthusian constraints have
been incorporated into mathematical models of population and economic
growth over the very long term (G. Steinmann and J. Komlos, 1988;
Komlos and M. Artzrouni, 1990).
The assessment of scale economies and technical change has not
been substantially modified in recent writings. It may be summarized
as follows:
(1) Economies of scale in manufacturing tend to be exhausted
at moderate market sizes; besides, it is not the size of
the internal market but the extent to which countries
are able to exploit the global market which determines
whether they can be reaped. The same can be said for
agglomeration and urbanization economies, which tend to
be exhausted after a country has one city of moderate
size. On both accounts, rapid population growth is
unlikely to be a stimulus to industrial growth in most
low-income countries.
(2) Economies of scale in agriculture and the provision of
rural infrastructure may be substantial. These arise,
however, from population density, not the rate of
population growth.
(3) Induced innovation and technical change are a major
force in the area of agriculture; V. Ruttan and Y.
Hayami (1991) cite studies in which their effects may
be seen over a time-span of some 20 years. Their
relevance for manufacturing and the modern sector in
general is less-well established.
The embodiment of technical change in capital or, due to
improving standards of training and education, in workers, has long
been known to favour rapid population growth via "vintage" logic: a
more-rapidly growing population will have a younger labour force; a
more-rapidly expanding capital stock will consist, on average, of
newer machines. D. Blanchet (1988a) has noted that since the rate
of depreciation typically accelerates þ an old capital stock
depreciates more rapidly than a young one þ conventional fixed-
depreciation-rate models somewhat over-estimate the capital-dilution
effect of rapid population growth. The benefits arising from
technical advancement as embodied in the labour force have been
formally investigated by E. van Imroff (1988), who added to a neo-
classical growth model the assumption that only new workers employ
new technologies.
VII. Rapid Population Growth and Externalities
P. Demeny (1986) stated flatly that the possible negative
external effect of family fertility decisions was the single most
important concern in the economics of population. It is the
importance or non-importance of externalities which determines
whether Governments should "go beyond family planning", i.e., take
policy action to discourage or encourage fertility.
There has been considerable progress in theoretical models of
household fertility (Nerlove, Razin and Sadka, 1987), notably work
in the "new new home economics" tradition þ the second "new" refer-
ring to the introduction of overlapping generations into the
household utility maximization model. Some studies have demonstrated
that externalities are unlikely to arise in areas where they might
have been expected; e.g., T. Srinivasan (1988) has argued that it is
unlikely that fertility decisions will give rise to externalities in
the area of old-age support and social-security systems. In other
areas where externalities do arise þ e.g., provision of public goods
such as national defence and social services such as education þ it
should be possible to correct for them with a tax or user fee.
It has long been recognized that the most intuitively appealing
of externalities arising from child-bearing þ that individual
fertility decisions affect the wages not only of the additional child
but of all other workers as well þ represents a logical fallacy
because the same birth which depresses the wage rate will raise the
rate of profit; i.e., the birth is Pareto-neutral. On the other
hand, intriguing and, sometimes, counter-intuitive potential problem
areas have been identified. For example, Becker (1988) argued that
unless parents are assured by legal or institutional measures of
transfer payments from their children in old age, they will under-
invest in their children's education.
Up to the appearance of the 1986 NAS report, the subject of
externalities represented something of a theoretical curiosity
because no attempt had been made to quantify them. Lee addressed
this problem in a series of papers (Lee, 1991; Miller and Lee,
1990). The exercise is made difficult by two factors: (1) the wide
range of uncertainty surrounding point-estimates of the externality
in each of the areas addressed; and, especially, (2) the fact that
estimates of the total externality are dominated by prodigious
external dis-economies calculated to arise from dilution of national
mineral resources, including oil, and public lands.
Not surprisingly in view of the latter, Lee calculated
significant negative externalities to child-bearing only in countries
with substantial mineral or land resources; as he pointed out (Miller
and Lee, 1990, p. 295), including environmental amenities in the
exercise would probably lead to the estimation of significant
negative externalities across the board. The value of externalities
arising in areas other than common-property resources tends to be
small although, as mentioned above, there is a wide range of
uncertainty around the point-estimates.
VIII.Rapid Population Growth, Income Distribution and Poverty
One of the strongest predictions of neo-classical theory is that
rapid demographic increase, by increasing the supply of labour
relative to the capital stock, will reduce the wage rate relative to
the rate of return to capital, thereby worsening the distribution of
income.
In a much-cited paper Lee (1980; cf. also 1987) found strong
evidence of this phenomenon in England and other European countries
between the 13th and 19th centuries. In fact, the evidence was much
too strong, as the average estimated elasticities of the real wage
with respect to population was -1.6, a result inconsistent with the
common assumption that the elasticity of substitution is around unity
(an elasticity on the order -0.5 would be expected). Lee skirted
this difficulty by proposing that the elasticity of substitution
between land and labour was, in fact, much less than unity in pre-
industrial Europe, and specifying a two-sector general equilibrium
model accordingly. Solution of the model yields an elasticity of
substitution of 0.16, far lower than other estimates.
Pooling time-series and cross-sectional data for six countries
between 1500 and 1800, D. Weir (1991) estimated a real wage-
population elasticity of -1.2: more reasonable than Lee's result but
still lower than expected. Thus, long-run historical data indicate
a very strong inverse relationship between the rate of population
growth and the welfare of the population. Whether this is Malthusian
in origin or arises from institutional arrangements, as Weir
suggests, is open to discussion.
Turning to poverty, the causal links between population growth
and absolute deprivation are not well understood. A large number of
stylized facts are known: historically, poor households have tended
to be small; currently, big households tend to be poor, but high-
status households still tend to have high fertility (M. Lipton,
1983). High-fertility poor households outnumber high-fertility elite
households. Thus, high fertility is concentrated among the poor and
tends, on compositional grounds alone, to worsen the distribution of
income.
On the other hand, research has not established a strong causal
link running from high fertility to poverty. G. Rodgers (1984, p.
169) concluded his survey of work in the area as follows:
"There seems to be one basic theme þ conclusion would be too
strong a word þ which recurs across many situations and levels
of analysis: that high fertility and high rates of population
growth tend to have adverse effects on the incidence and
evolution of poverty, but that these effects tend to be
relatively small."
Birdsall and Griffin (1988) found evidence that poverty
encourages large family size, but wrote (p. 49), "Regarding the link
in the other direction þ the impact of rapid population growth on
poverty þ there is some theory but much less evidence." They
concluded, nonetheless, that slower population growth would reduce
the cost and time necessary to eliminate poverty.
IX. Correlation Studies and Other Models
Lee (1983, p. 54) cautioned that correlation studies need to
take into account dynamic aspects or else they will be inaccurate:
"If national economies followed the trajectories described by the
neo-classical single-sector growth models, then we would expect, for
steady-state economies, no relation between population growth rates
and growth rates of per capita income. If such economies are out of
steady state, then we would expect a moderately strong negative
correlation between them; and if the economies had surplus labour,
we would expect a strong negative correlation."
The correlation between the rate of population growth and the
rate of growth of per capita income was for a long time notorious as
a non-relationship. That is, when researchers compared the
population growth rates of countries with their per capita economic
growth rates, either in informal graphical and tabular presentations
or in formal correlation analyses, there was generally no discernable
relationship. On the other hand, more recent correlation studies
have inferred a significant inverse relationship for the 1970s and
early 1980s (Blanchet, 1991b), although not for earlier years and not
always for the post-war decades combined.
The simplest explanation for this revolves around the second of
Lee's possibilities. During the "golden age" of the 1950s and 1960s,
benign economic conditions allowed countries to accommodate rapidly
growing populations along the lines suggested by the mainstream neo-
classical model. In the 1970s and early 1980s, by contrast,
economies were displaced from their growth paths by exogenous shocks,
in particular swings in primary commodity and energy prices and the
emergence of the debt crisis. Under ideal conditions, neo-classical
substitution processes would quickly return displaced economies to
their equilibrium growth paths, but these processes were hampered by
institutional rigidities which resulted in incomplete adjustment and
attendant unemployment, inflation, price distortions, protectionism,
etc. This in itself would be sufficient to lead to the emergence of
a negative correlation; it might furthermore be argued that countries
with rapid population growth found it more difficult to adjust than
did those with moderate demographic increase.
Part of the observed change may also be spurious. Say, for
example, that aggregate GDP growth rates remained unchanged, while
population growth rates fell in countries already characterized by
relatively slow population growth and remained fixed in countries
experiencing relatively rapid rates of demographic increase. Then
whatever the underlying model or absence thereof, the slope co-
efficient estimated by regressing the rate of per capita income
growth on that of population growth would become more negative in
arithmetic value.
An alternative explanation is that there actually is an
underlying "Malthusian" model which correlation analysis is only now
starting to reveal. Blanchet (1988b) has advanced an econometric
argument in furtherance of this view. The basic Malthusian model
should, according to Blanchet, consist of two equations and an
identity þ the rate of per capita GDP growth as a behavioural
function of the rate of population growth, the rate of population
growth as a behavioural function of the level of income, and the
level of income as last-year's level multiplied by one plus the
growth rate, an accounting identity. Blanchet argues that the
underlying model is of the error-components variety (R. Pindyck and
D. Rubinfeld, 1976, pp. 202-08). Simulation analysis with reasonable
parameter estimates shows that ordinary least squares estimates of
b will "cycle": Blanchet (1988b) illustrates the case where they
start off positive and insignificant for decade one, grow steadily
more positive through decade three, then suddenly turn negative and
significant in decade four.
Kelley (1986) expressed hope that the NAS report would stimulate
more activity in the area of multisectoral population-development
modelling, a hope which has not been borne out. Still worthy of
attention, however, is the model of Wheeler (1984), which is perhaps
the most credible simultaneous model of population growth, investment
and human-resource quality as measured by nutrition, literacy and
life expectancy.
The Wheeler model stands out because it is characterized not
only by strong simultaneous linkages þ for example, rising levels of
health raise per capita income, which in turn has a beneficial
feedback effect on health þ but also, and less commonly, by
econometric credibility in terms of the estimation techniques
employed. While finding that reduction of fertility via family
planning was a highly effective development policy, simulations with
the Wheeler model indicated that given even minimal investments in
education and family-planning, low-income countries were unlikely to
remain mired in a "low-level trap".
R. Barlow (1992) has estimated a single-equation reduced-form
model in which the growth rate of GDP is a function of current-period
and lagged birth rates, plus a range of variables reflecting the
external environment and socio-political factors. Stripped to its
essentials, the model is
Y'i(t) = a + b NBRi(t) + c NBRi(t-1) + d NBRi(t-2)
+ e NBRi(t-3) + Other variables + fi(t)
where
Y'i(t) = Average annual GDP growth rate in country i in period
t
NBRi(t) = Net birth rate (births minus infant deaths) in country
i in period t
Other variables = improvements in life expectancy, political
stability, terms of trade, etc.
fi(t) = country-specific random errors.
and the periods correspond to six-year intervals. The coefficients
on current and lagged fertility are hypothesized to reflect both
labour-force growth (negative in the current period because of
reduced female labour-force participation, positive for lagged
periods) and investment (negative in the current and near-lag periods
because of the household dependency-rate effect, positive in far-lag
periods because of the higher ratio of workers to retirees).
When estimated over a sample of developing and industrialized
countries observed between the mid-1970s and early 1980s,
coefficients indicate a large negative current-period fertility
effect and a positive lagged fertility effect. The current-period
elasticity of output with respect to the net birth rate is about
-3.0; after six years it is -1.5, after 12 years -0.2 and after 18
years 1.2. In one simulation comparing a country with a total
fertility rate (TFR) of 5.0 versus the same country with a TFR of
3.5, after 24 years GDP is 62 per cent higher, population 18 per cent
lower and per capita GDP a staggering 99 per cent higher in the low-
fertility scenario.
The outlines of the story told by the Barlow model may be
correct, but there is reason to believe that the estimated effects
may be too strong. Parameter estimates reflect not only differences
in actual fertility behaviour, but differences in the age structures
of populations. For example, the population of a country with a net
birth rate of, say, 15 per 1,000 is characterized by a lower
dependency ratio than a country with a net birth rate of, say, 40 per
1,000. There is evidence (some of which we have discussed above in
the context of saving rates) that the less-dependent age structure
may be more conducive to economic growth; arguments involving the
age-structure of the labour force may also be advanced.
Two other comments may be made regarding the Barlow model.
First, omitted variables are an ever-present difficulty in work of
this sort, and Barlow has included a wide range of socio-political
variables to mitigate the problem. Based on his parameter estimates,
the rapid growth of the East Asian "Dragons" should be attributed
mostly to their rapid fertility decline. Omitted variables may have
led to an overstating of the correlation in these cases. Second, the
results of the simulation exercise presented must be interpreted with
caution. For example, included among the ceteris paribus assumptions
is that a country with a total fertility rate of 5.0 and one with a
total fertility rate of 3.5 start off with identical age distri-
butions.
X. Conclusion
This survey has found a number of areas in which recent research
has provided further evidence that rapid population growth has a
negative impact on economic development (see summary below). Parti-
cularly noteworthy population-growth effects are the slower
absorption of labour into high-productivity sectors, and the adverse
consequences at the household level for child welfare and human-
capital formation. Recent research has found, however, that the
magnitudes of the latter effects are modest, and more work is
required to assess their long-run impacts. Further work is also
necessary to quantify the macroeconomic effects of strains placed by
rapidly growing populations on national health and education
resources.
Research in the areas of savings and investment and poverty has
not been conclusive, and little new work has been done in the areas
of scale economies and induced technical and institutional change.
Recent work has suggested the existence of large external dis-
economies to child-bearing in the area of common-property natural
resources in countries where these are substantial; the value of
externalities in other areas has been estimated to be small. The
existence of a strong negative correlation between wages and
population growth in pre-industrial Europe has been reaffirmed, and
recent work has found a statistically significant negative
correlation between the rate of economic growth and the rate of
population growth during the 1970s. Both correlations are open to
different interpretations.
In all, the recent research surveyed in this paper does not call
for substantial modification of the following summing up of the U.S.
National Academy of Sciences review:
(1) Rapid population growth has a negative effect on
development in many developing countries.
(2) This role is difficult to quantify due to the complexity
and multiplicity of relationships involved and the
variability of local circumstances.
(3) Because there exist possibilities for response and
substitution, the effect of population growth on
economic development is probably a modest one.
Nevertheless, recent evidence indicates that point (3) may be
too mild an appraisal.
In the years since 1986, population policy makers, including
policy advocates in international development institutions, have
become more concerned about the adverse effects of rapid population
growth on economic development. This trend, as has been seen, cannot
be explained by a pronounced shift in the results of economic
research in the area over the same period. In the face of the
uncertainties of research findings, it appears that policy makers
have acted in a prudent manner to implement population policies to
slow growth.